Weird: HF dip when L+R are driven, flat when just L or R is driven

[Chaconne]

From the fun-with-iPhone-RTAs department: I was just messing around with the RTA app on my phone, playing pink noise. I noticed a dip of about 5 Db in the high frequencies centered at about 9K and stretching in a U shape from about 6K to 12K. I decided to use my balance control (remember those?) to check the response of just the left channel. When I did, the dip completely disappeared. The same thing happened when I played the right channel alone. I re-centered the balance, and back came the very same dip. I did this several times with the same results.

The phone was at one point of an equilateral triangle formed with the two speakers, about 5-6 feet from each speaker, with each speaker's axis aimed at the phone's mic. Anyone know why this happened? I know a little bit about frequency response, room interaction, etc., but this one puzzles me.

 

[Blumlein 88]

Two sources separated by a distance will have points of cancellation. Your results fit for a difference in distance between the two speakers of some inch or two. When you drop the signal from one channel then there is no cancellation null.

Basically an example of comb filtering.

 

[Chaconne]

Thanks, Blumlein. The thought of cancellation crossed my mind, but I'm so used to thinking of that in terms only of bass frequencies that I blew it off. But I guess sound waves are sound waves, regardless of frequency!

 

[mhardy6647]

I deleted this comment, but I'm going to reanimate it

Two sources separated by a distance will have points of cancellation. Your results fit for a difference in distance between the two speakers of some inch or two. When you drop the signal from one channel then there is no cancellation null.

Wouldn't the nodes and antinodes be pretty close together for treble? Lambda for 5 kHz is about 25 cm.
EDIT: Like heck it is! Thanks @Blumlein 88 ! See post(s) below). Mea culpa. I am chagrined.

I am wondering if perhaps the treble crossover in one loudspeaker is wired with reversed polarity with respect to the other loudspeaker's? Such QC gaffes do occur.
I would think some difference in FR for one loudspeaker assesed in mono compared to the other (also mono) would manifest itself, but I'd certainly expect to see a more blatant impact if both were fed a mono signal and played at the same time.

 

[Blumlein 88]

I deleted this comment, but I'm going to reanimate it



Wouldn't the nodes and antinodes be pretty close together for treble? Lambda for 5 kHz is about 25 cm.

I am wondering if perhaps the treble crossover in one loudspeaker is wired with reversed polarity with respect to the other loudspeaker's? Such QC gaffes do occur.
I would think some difference in FR for one loudspeaker assesed in mono compared to the other (also mono) would manifest itself, but I'd certainly expect to see a more blatant impact if both were fed a mono signal and played at the same time.

I think it is more like 2.5 cm or so at 5 khz. Remember a half wavelength difference will cancel. Even if you have your speakers arranged so that you think they are exactly equidistant it isn't hard to be off by an amount of 1 to 2.5 cm. It also could be part of a hard reflection on one side vs the other. You could try moving the microphone doing the measuring a small amount side to side to see if you find a spot where the treble dip goes away. Remember the OP said it was centered around 9 khz and seemed to extend from 6-12 khz. All of that sounds like a cancellation from differening distance. Also it was a 5 db dip. If polarity was reversed on one it might be a deeper dip.

 

[radix]

I think it is more like 2.5 cm or so at 5 khz. Remember a half wavelength difference will cancel. Even if you have your speakers arranged so that you think they are exactly equidistant it isn't hard to be off by an amount of 1 to 2.5 cm. It also could be part of a hard reflection on one side vs the other. You could try moving the microphone doing the measuring a small amount side to side to see if you find a spot where the treble dip goes away. Remember the OP said it was centered around 9 khz and seemed to extend from 6-12 khz. All of that sounds like a cancellation from differening distance. Also it was a 5 db dip. If polarity was reversed on one it might be a deeper dip.

If you use the REW generator and play a single tone, e.g. 5 kHz, it is easy to move your head around the room and you will clearly hear the deep nulls.

I've done this before, but I think I was more like 800 Hz. 5 kHz might be harder to hear due to the smaller wavlength.

 

[DonH56]

Comb filter effect: https://www.audiosciencereview.com/forum/index.php?threads/comb-filter-effects.25432/

 

[DVDdoug]

It's been awhile since I played around with pink or white nose but you can probably hear different frequencies going in-and-out of phase when you move around. You WILL hear loudness changes with pure test tones (at maybe 5kHz and higher). You'll probably get that with one speaker due to the sound bouncing around the room.

We don't usually notice it with music. I assume because music is complex and we are used to it. I was doing an experiment with test tones and an SPL meter once and not only was I hearing the tones going in-and-out of phase but just moving around behind the SPL meter made the readings change by a few dB.

 

[Chaconne]

Thanks to everyone for the responses. Lots of food for thought.

 

[dasdoing]

dip of about 5 Db in the high frequencies centered at about 9K and stretching in a U shape from about 6K to 12K

it's obviously cancelation. but I doubt it is the normal stereo effect other are sugesting. if it was drastic like this stereo would never had success.
it either a problem of the crossovers beeing diferent in the speakers, or there is a strong reflection happening on one side only.

 

[mhardy6647]

I think it is more like 2.5 cm or so at 5 khz. Remember a half wavelength difference will cancel. Even if you have your speakers arranged so that you think they are exactly equidistant it isn't hard to be off by an amount of 1 to 2.5 cm. It also could be part of a hard reflection on one side vs the other. You could try moving the microphone doing the measuring a small amount side to side to see if you find a spot where the treble dip goes away. Remember the OP said it was centered around 9 khz and seemed to extend from 6-12 khz. All of that sounds like a cancellation from differening distance. Also it was a 5 db dip. If polarity was reversed on one it might be a deeper dip.

Did I slip a decimal point?
lemme check. speed of sound = 34,300 cm/sec per google
I checked it 1234.8 km/hr = 34,300 cm/sec yup
5 kHz <=> 0.0002 sec per ahem cycle.
0.0002 sec * 34,300 cm/sec = 6.86 cm wavelength
Yup, sorry, thanks! Serves me right not to do the arithmetic myself (I trusted google, silly me)...

But, yeah, my point exactly.... and testable (although the capture angle of the measuring mic might make nodes - or antinodes - hard to 'see').

Thanks (and, again, sorry)!!!

PS FWIW I agree with @DVDdoug -- simply movin' around the room a bit, bobbin' and weavin', should manifest some interesting aural effects, especially if done with a fixed tone (say... oh, I dunno... 5kHz ).

 

[fpitas]

See? This is why mono was better!

/

 

[youngho]

From the fun-with-iPhone-RTAs department: I was just messing around with the RTA app on my phone, playing pink noise. I noticed a dip of about 5 Db in the high frequencies centered at about 9K and stretching in a U shape from about 6K to 12K. I decided to use my balance control (remember those?) to check the response of just the left channel. When I did, the dip completely disappeared. The same thing happened when I played the right channel alone. I re-centered the balance, and back came the very same dip. I did this several times with the same results.

The phone was at one point of an equilateral triangle formed with the two speakers, about 5-6 feet from each speaker, with each speaker's axis aimed at the phone's mic. Anyone know why this happened? I know a little bit about frequency response, room interaction, etc., but this one puzzles me.

The speed of sound is about 1130 ft/s. 9 kHz is about 0.1255 ft for wavelength or about 1.5 inches. Half of a 9 kHz wavelength would then be 0.75 inches, so that amount of difference in terms of distance between one speaker and the other relative to the measurement position could result in destructive interference at 9 kHz due to the sound from each speaker arriving out of phase with respect to each other.

 

[mhardy6647]

See? This is why mono was better!

/

or maybe googlephonics, which I believe ended up being called Dolby Atmos, probably because of some TM issue with Alphabet.

(N.B. Steve's shtick uses a Bad Word... several times, in case any of all y'all are... you know... triggered by such things. )